$ figures are API-equivalent compute cost (cache-aware pricing). On a subscription, marginal $ is 0 — read as a compute-burn meter. TRUE $ adds the human attention you spend.
RANGE
AGENT
SOURCE
PROJECTMODEL
How you compare your 30-day cohort vs the field · percentiles over contributing orgs (k-anonymous) · aggregates only — see PRIVACY.md
loading…
Model Economics sorted by TRUE $/COMMIT = (compute $ + prompts × attention $) per shipped commit — lower wins · click a model to filter
Daily Burn API-equivalent $ per day, stacked by model family
Sessions click a row for detail · recaps are AI-generated (haiku)
Reports generated by `codenomics report weekly|monthly`
Economic drivers
Your inputs to the canned metrics. Saved to ~/.config/codenomics/config.json and applied everywhere (dashboard + reports) instantly.
what one interactive prompt of your attention is worth (context switch + review). Machine sessions: $0.
optional loaded rate; adds active supervision time of human sessions to TRUE $. 0 = off.
Limits
Budgets evaluated per calendar window. Breaches show here and fail codenomics index --check-budgets.